The Economics of Free and Open Source Software

An economic model of the profit-seeking firm seeking to deploy "Open Source" or "Free" software. Both firms considering "purchase" and development of such software are considered. Open Source software as a public good. Such software is a quasi-public good which will sometimes, but not always, The last fifty years have seen software evolve from an initial status of something given away freely with the purchase of hardware to a profitable business as something salable in its own right, and then again to something given away without charge on its own, without any sale of underlying hardware. At first impression, it would appear easier to explain why firms would use such software than why they would produce it, but upon building appropriate economic models of production, it is easier in many cases to explain the production than the adoption of such software. When adopting software for business purposes, the purchase price is only a minor consideration. The appropriate cost for the profit seeking firm is not merely the purchase price, but the support costs of the software as well, which is typically far higher than the software cost. There will frequently not be a firm selling such software, and the must either purchase outside support (if available), rely on its internal support, or rely on "community support" from other users or the developers. Depending upon the quality of the software, this may make the "free" software either more or less expensive than commercial software. The popularity of such programs as sendmail (which is involved in the transmission of most email messages), Apache (by far the most popular web serving software) and the Linux operating system suggest that this condition can be met in the commercial world. It is for the developers of software that there is the potential to significantly reduce costs or enhance revenues. While developing software under an open source license raises the possibilities that a competitor will use the the work in competition to the developer, that people will take the software without paying, or both, it does not appear that the firms choosing such a route are in the business of selling software. Instead, such firms need such software to exist in order to achieve their primary mission, and find open source solutions to be lower cost. For example, IBM and Sun sell powerful computers to function as web servers. There are not separate demands for hardware and server software, but instead a demand for pages served. If either company can reduce its costs, it can gain an advantage in this market. While IBM previously had a proprietary product, it now uses Apache, and also contributes resources to the ongoing development of Apache. Sun also uses Apache, but has gone a step further than IBM. Sun purchased the maker of StarOffice, a competitor to Microsoft Office, and gives the program away in an attempt to sell more servers. Additionally, Sun has published the source code under an open license, in the hopes of selling more servers by having a better "product" and to gain outside assistance. IBM competes with other users of Apache, and therefore with the improvements that it makes in Apache. Sun faces the possibility of the code to StarOffice being used in a competing product. Nonetheless, both continue to expend resources on development of the products, in expectation of higher profits. Development costs are potentially lower for a variety of reasons: the ability to use code that has already been written elsewhere, the fact that outsiders continue to develop the code, and that outsiders reduce the number of bugs in the code, lowering support costs. In addition to a potentially lower cost of production, the firm is still able to charge customers for support of the product--either as a purchase price or with a separate contract. If the firm has particular expertise with the software, it is likely that the firm can charge a greater price than another firm could for the support of the same software, meaning that the firm will retain a competitive advantage. The paper develops an economic model for such decisions, weighing the potential losses from the development of open source with the potential gains, thus showing the conditions under which developing under an open source license is the profit maximizing decision. It considers two types of licenses, those in which a later party may make changes and not disclose the changed source code, such as the BSD and MIT licenses, and those in which changes must be disclosed, such as the GPL. Finally, open source is analyzed as a quasi-public good, in which it is possible that either a single entity or a small number of entities may receive beneifits greater than the cost of providing the good.